Health Disparities
Assessing Subjective and Objective Cognitive Function Across Intersectional Identities Using Multilevel Analysis of Individual Heterogeneity and Discriminatory Accuracy (MAIHDA) Nancy Chen* Nancy Chen Chen Chen Chen Chen Chen Chen Chen Chen Chen University of California, Davis
The extent to which intersecting social identities contribute to disparities in subjective and objective cognitive performance remains poorly quantified. We applied an intersectional framework to examine differences in subjective and objective cognition across identities of gender, race/ethnicity, and education.
We analyzed 20,814 participants aged 45+ from the Brain Health Registry. Subjective cognition was measured using the 39-item Everyday Cognition (ECog) Scale, operationalized as the mean score across items. Objective cognition was measured using the Cambridge Cognition Paired Associates Learning test including First Attempt Memory Score (FAMS) and Total Errors Adjusted (TEA). We defined 24 intersectional strata based on gender (men, women), race/ethnicity (Latino, Asian, Black, White), and education (≤2-yr degree, 4-yr degree, Master’s/Doctoral/Professional degree). Using the multilevel analysis of individual heterogeneity and discriminatory accuracy (MAIHDA), we fit multilevel linear models with individuals nested within intersectional strata. We estimated age-adjusted null models and main effects models additionally adjusted for gender, race/ethnicity, and education, and quantified between-stratum variance (VPC) and proportional change in variance (PCV).
Mean age was 67±9 years. The most frequent intersectional stratum was White women with a Master’s/Doctoral/Professional degree. The variance between intersectional strata accounted for ≤6% of the total variance in ECog, FAMS, and TEA. Most (79-98%) of this between-stratum variance was explained by additive effects of gender, race/ethnicity, and education.
The variance in subjective and objective cognitive performance at the intersectional strata level was small and mostly explained by the additive, not interactive, effects of gender, race/ethnicity, and education. Carefully specified models that adjust for these social factors may adequately capture major disparities in cognitive outcomes.
